×

Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

作者: Will Ma ×
换一批
01.
arXiv (CS.CL) 2026-06-17

A Framework for Evaluating Agentic Skills at Scale

Agent skills – structured, reusable knowledge artifacts that augment LLM agent capabilities – have been rapidly adopted in industry, yet their cross-domain impact and use across commercial and open-source models remain under-studied, and no reusable methodology exists for evaluating an individual skill. In this work, we present an evaluation framework that lets a skill author construct realistic tasks to rigorously assess the aspects of a skill that matter most to them, and that estimates skill utility by solving those tasks. Further, we apply our evaluation approach at scale to 500 real-world skills, generating 1,000 tasks derived from the skills' content, along with instruction-following and goal-completion scoring rubrics. Using these metrics, we evaluate how 19 agent-model configurations, both proprietary and open-source, perform on the tasks. Our results show that models vary widely in how closely they adhere to the instructions encoded in skills, leading to substantial differences in their performance gains. Furthermore, we show that access to a skill significantly changes model behavior compared to the no-skill setup, providing an essential mechanism for encoding opinionated workflows into LLM agents. We release our evaluation dataset to support future work on agent skills.

02.
arXiv (CS.AI) 2026-06-19

Which Pairs to Compare for LLM Post-Training?

arXiv:2606.19607v1 Announce Type: new Abstract: Preference-based post-training has become a central paradigm for aligning language models. A common data-collection strategy is to generate a small set of completions for each prompt and label the resulting comparison pairs. However, human preference labels are often much more expensive than generating additional completions, suggesting a different use of the same labeling budget: generate a larger pool of completions, but label only the most informative comparison pairs. This paper studies which pairs should be compared in preference-based post-training. We formulate comparison curation as a sampling-design problem and evaluate designs by the quality of the final policy under the preference-based post-training objective. We instantiate this framework for Direct Preference Optimization (DPO), analyzing how the choice of labeled pairs propagates through DPO training to downstream policy performance. Our main results provide matching upper and lower bounds on the post-training optimality gap of the DPO-trained policy. The bounds show that comparison selection affects downstream performance through a single design-dependent information matrix, which links label allocation to parameter estimation error and policy suboptimality. This yields an explicit optimization criterion for budgeted comparison curation and motivates practical sampling designs for selecting informative pairs from large generated completion pools. Experiments on synthetic settings and language-model post-training benchmarks show that the proposed designs consistently improve sample efficiency over common comparison-selection heuristics.

03.
arXiv (CS.LG) 2026-06-16

Towards Data-Efficient Cross-Device Generalization of Grad-Shafranov Equilibria via Transfer Learning Neural Operator

arXiv:2606.15512v1 Announce Type: new Abstract: Real-time reconstruction of magnetohydrodynamic equilibria is essential for plasma shaping, stability assessment and feedback control in magnetic confinement fusion. However, Grad-Shafranov equilibrium calculations remain largely device-specific and iterative, limiting their use in latency-constrained control settings. Existing neural approaches can accelerate individual equilibrium predictions, but they do not generally provide reusable models across changing plasma boundaries or tokamak geometries. Here we show that equilibrium reconstruction can be recast as a cross-device operator learning problem. We develop a domain-specific neural operator framework that maps geometry and profile parameters directly to the poloidal flux field, replacing repeated solve-on-demand computation with amortized operator inference. Using the analytically tractable Solov'ev family as a controlled Grad-Shafranov testbed, we generate equilibria across eight geometrically distinct tokamak-like configurations and benchmark five neural operator architectures under four transfer-learning strategies. Single-geometry pretraining gives poor transfer to unseen devices, whereas multi-geometry pretraining enables data-efficient adaptation. The Wavelet Neural Operator gives the strongest cross-geometry performance, reaching mean relative L2 errors below 4% with 100 labelled target equilibria and below 2% with full fine-tuning. The predicted magnetic fields satisfy the divergence-free constraint to numerical precision, and four architectures achieve millisecond or sub-millisecond inference. These results identify neural operator pretraining as a route towards reusable, real-time equilibrium inference across fusion device configurations.

04.
arXiv (CS.CV) 2026-06-11

How Seemingly Inconsequential Design Choices Dictate Performance of LLMs in Pathology

General-purpose large language models (LLMs) are routinely used as baselines when evaluating specialized pathology models on whole-slide images (WSIs). Because WSIs exceed contemporary model context limits, LLM baselines routinely use small, high-magnification patches processed independently via majority voting, without systematic evaluation of seemingly inconsequential design choices such as patch size, patch count, and magnification. Generalist LLMs have consistently underperformed specialized systems, reinforcing the perception that domain-specific training or architectural adaptation is necessary for pathology tasks involving WSIs. Here, we conduct a systematic factorial analysis of four input design factors: inference mode, patch size, magnification, and patch count. We demonstrate that prior studies have overstated the gap between specialized models and general-purpose LLMs by choosing non-optimized input configurations. On the MultiPathQA benchmark, switching to a single balanced configuration (large patches at lower magnification, processed jointly) raises GPT-5 from 15.1% to 39.5% on cancer-type classification (TCGA) and from 38.1% to 62.9% on organ classification (GTEx). Per-task optimization yields further gains up to 43.9% (TCGA) and 71.6% (GTEx). The same configuration generalizes to two other models and to a fully held-out CPTAC cohort, where it improves Gemini 3 Flash by 23.4 percentage points without any task-specific tuning.

05.
medRxiv (Medicine) 2026-06-17

Identifying anaphylaxis using weakly-supervised prediction models and natural language processing

Objectives Scalable computable phenotyping algorithms are critical for conducting high-throughput disease-outcome research in large, distributed-data electronic health record (EHR) and claims data settings. We developed and evaluated a claims- and EHR-based computable phenotyping algorithm for anaphylaxis, a rare acute condition that is challenging to accurately identify using claims data alone. Materials and Methods Potential anaphylaxis events came from two healthcare systems (Kaiser Permanente Washington [KPWA] and Vanderbilt University Medical Center [VUMC]). We engineered features from clinical text using automated natural language processing (NLP) methods. We then developed a phenotyping algorithm using four NLP- and diagnosis code-based silver labels (proxies for the gold-standard labels). Gold-standard abstracted outcomes were used to evaluate algorithm performance. Results The largest area under the receiver operating characteristic curve (AUC) was 0.931 for an NLP-based silver-label model at KPWA. Depending on the model and healthcare system site, positive predictive value (PPV) and sensitivity at the threshold of predicted probability that maximized F1 score ranged from 0.52 to 0.77 (PPV) and 0.78 to 1 (sensitivity). Discussion NLP-based silver-label models had large AUC at KPWA but not at VUMC. This may be because clinical text at KPWA is only available for outpatient encounters and secure messaging. High sensitivity for identifying anaphylaxis can be obtained using our best-performing models. Conclusion The best-performing models had better PPV and sensitivity tradeoffs than prior bespoke anaphylaxis models with costly, manually curated features. The simplicity of the approach compared to traditional phenotyping methods allows it to be deployed easily at multiple health care systems.

06.
medRxiv (Medicine) 2026-06-11

Computer Vision for Real-Time Anatomical Navigation in Neurosurgery: First-in-Human Clinical Evaluation and Iterative Development (IDEAL Stage 1)

Introduction: Precise anatomical navigation is fundamental to safe endoscopic pituitary surgery, a high-stakes procedure characterised by a challenging learning curve. While traditional navigation systems often rely on workflow-disrupting probes or static preoperative imaging, advancements in computer vision AI (CVAI) now enable dynamic, real-time anatomical segmentation directly from live surgical video1-3. Our group has previously conducted a series of preclinical human-computer interaction studies to refine the system's design, alongside digital and high-fidelity physical simulations demonstrating the benefit of AI assistance in improving overall performance, training, and safety4-8. Building on this foundation, the current study represents a first-in-human application of real-time CVAI assistance in the neurosurgical operating room, serving to assess feasibility and safety, and to iteratively improve the system. Method: Guided by DECIDE-AI and IDEAL frameworks, this single-centre evaluation comprises an initial proof-of-concept phase (n=6) for endoscopic transsphenoidal pituitary surgeries. The AI model utilised a DINOv3-derived vision transformer architecture, deployed via a high-performance edge computing unit to achieve low-latency, real-time inference without reliance on cloud infrastructure2. Given the high-risk nature of the procedure and the early stage of clinical AI integration, the system was initially deployed as an educational adjunct on a secondary monitor, ensuring the primary surgical feed remains uncompromised. Functionality and safety were assessed via structured questionnaire, prospective observation, and blinded retrospective review of the recordings of the endoscopic surgical video feed and wider operating room environment. Continuous multi-stakeholder feedback through validated human factors surveys drove iterative technical refinements between cases. Results: Six patients with pituitary adenomas were enrolled. The CVAI system was successfully deployed in four cases, demonstrating acceptable real-time sella segmentation accuracy. Deployment failed pre-operatively in two cases owing to a single recurring system reboot bug. Iterative refinement between cases were driven by our experience and surgical team feedback. This resulted in the integration of additional anatomical structure segmentations (e.g., carotid arteries), enhanced model accuracy via training dataset expansion, and hardware firmware upgrades. Multi-stakeholder surveys demonstrated satisfactory system feasibility, usability, and acceptability among the surgical team. Both prospective observation and retrospective video review confirmed the absence of adverse events, including no significant distraction to the primary surgeon, and there were no AI-related clinical complications. Conclusion: This first-in-human early clinical evaluation demonstrates the feasibility, safety and iterative development of real-time, CVAI-based anatomical navigation during high-stakes neurosurgery. Future work will include a larger single-centre case series (IDEAL Stage 2a) with more surgical teams to further iterate the system and explore its impact on training and workflow. As the underpinning technology improves, deployment will transition to direct intra-operative decision support and integration with other intra-operative navigational technologies.

07.
arXiv (quant-ph) 2026-06-19

Application and quantum properties of superpositions of oppositely squeezed states

arXiv:2511.03204v2 Announce Type: replace Abstract: We show that superpositions of oppositely squeezed states – non-Gaussian Schr{\"{o}}dinger-cat-like states – exhibit enhanced nonclassical features and provide an entanglement advantage in the small-squeezing regime. These states possess photon-number structures distinct from conventional coherent-state cat states, and we analyze their Wigner functions and the entanglement generated when they are injected into a 50-50 beam splitter. As a practical application, we demonstrate that they enable a high-quality heralded single-photon source whose second-order intensity correlation function is smaller than that obtained from a pure two-mode squeezed vacuum state. We further propose a linear-optical heralding scheme that approximates these superpositions without requiring strong Kerr nonlinearities. Our results indicate that the superposition of oppositely squeezed states is a promising non-Gaussian resource for quantum information processing, particularly for single-photon generation.

08.
arXiv (CS.LG) 2026-06-19

On the Oracle Complexity of Interpolation-Based Gradient Descent

arXiv:2606.19878v1 Announce Type: new Abstract: Recent work on first-order optimizers for empirical risk minimization (ERM) has suggested that smoothness of ERM loss functions in the training data, rather than in the optimization parameters, can be leveraged to improve the oracle complexity of gradient descent (GD) methods. In this paper, we propose an inexact gradient method, piecewise polynomial interpolation-based gradient descent (PPI-GD), which approximates the full gradient in each iteration by querying the first-order oracle at equidistant points in the data domain to construct polynomial interpolants of the resulting gradient samples over appropriately sized patches of the data domain. We analyze the oracle complexity of PPI-GD for strongly convex and non-convex loss functions when the data space dimension is bounded by a polylogarithmic function of the number of training samples, and find it to outperform several GD variants in key regimes when the loss function is sufficiently smooth. Furthermore, our analysis extends several techniques from the error analysis of bicubic spline interpolants to the setting of $d$-variate tensor product polynomial interpolants which may be of independent interest in interpolation analysis.

09.
arXiv (CS.LG) 2026-06-15

Running the Gauntlet: Re-evaluating the Capabilities of Agents Beyond Familiar Environments

arXiv:2606.14397v1 Announce Type: new Abstract: As agentic systems continue to evolve and are widely deployed in real-world scenarios, there is a growing demand to faithfully evaluate their capabilities. However, current benchmarks are typically built on popular applications with relatively simple tasks and focus on a narrow set of capabilities while overlooking broader dimensions, resulting in saturated performance on modern agents and failing to probe their limitations. To this end, we introduce GauntletBench, a web-based benchmark for evaluating agent generalisation in challenging scenarios, focusing on three underexplored capabilities (temporal perception, graphical understanding, and 3D reasoning), across five less-covered professional applications (Video Editor, Workflow Builder, 3D Modeller, Flight Analyser, and Circuit Designer), each with 20 vision-intensive tasks (100 in total). Our benchmark provides a modular pipeline that comprises an environment compatible with both open- and closed-source agent frameworks, a controlled web-based application, a well-structured task suite, and an automated evaluation engine with diverse metrics. Contrary to widespread expectations, our empirical results reveal that frontier agentic systems remain far from achieving human-level performance. Even the state-of-the-art agent achieves only a 19.1% success rate on our GauntletBench, highlighting the limitations in these overlooked capabilities and generalisation. By comparison, non-expert human annotators achieve over 80% success on our challenging yet feasible tasks, revealing the substantial gap between current agent capabilities and those required for complex real-world scenarios.

10.
arXiv (CS.CL) 2026-06-16

Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

We introduce Nemotron 3 Ultra, a 550 billion total and 55 billion active parameter Mixture-of-Experts Hybrid Mamba-Attention language model. We pre-trained Nemotron 3 Ultra on 20 trillion text tokens, then extended the context length to 1M tokens, and post-trained using Supervised Fine Tuning (SFT), Reinforcement Learning (RL), and Multi-teacher On-Policy Distillation (MOPD). Nemotron 3 Ultra is our most capable model yet, employing multiple key technologies - LatentMoE, Multi Token Prediction (MTP), NVFP4 pre-training, multi-environment RLVR, MOPD, and reasoning budget control. Nemotron 3 Ultra achieves up to ~6x higher inference throughput as compared to state-of-the-art publicly available LLMs while attaining on-par accuracy. The state-of-the-art accuracy, high inference throughput, and 1M token context length make Nemotron 3 Ultra ideal for long-running autonomous agentic tasks. We open-source the base, post-trained, and quantized checkpoints, along with the training data and recipe on HuggingFace.

11.
medRxiv (Medicine) 2026-06-18

Hospital staff views on the visibility, role and impact of Acute Learning Disability Liaison Services in Wales: a service evaluation

People with a learning disability experience marked health inequalities. In Wales, Acute Learning Disability Liaison Services (ALDLS) are delivered by specialised learning disability services, and all roles within them are undertaken by Learning Disability Liaison Nurses (LDLN). These services aim to enable access to, and delivery of, secondary care by supporting reasonable adjustments, facilitating communication, and coordinating care for people with learning disability during hospital encounters. However, independent evidence of the impact of ALDLS on patient care remains limited. This evaluation tries to address this evidence gap by examining hospital staff perceptions of the visibility, role, and impact of ALDLS across Welsh Health Boards, with the aim of informing service design and development and improving secondary care access and care for people with learning disability. The service evaluation used a qualitative approach involving interviews and a focus group with hospital staff across the seven Welsh Health Boards who had experience working with or interacting with ALDLS staff to care for patients with learning disability. Findings cover six key areas including i) visibility and delivery of ALDLS, ii) Barriers and challenges to effective ALDLS delivery, iii) Enablers of effective ALDLS delivery, iv) Positive impacts for patients with learning disability, v) Negative impacts and unintended consequences when the service is absent or limited, and vi) Participants recommendations for future improvements of ALDLS. To synthesise the findings, we developed an overview diagram, which illustrates how ALDLS may influence care quality in acute hospitals. The overview places the liaison service at the centre, showing how organisational enablers and barriers shape its delivery, and how its core functions support improvements in safety, timeliness, effectiveness, efficiency, equity, and patient-centred care. From the findings we have identified recommendations for practice and policy. These include that ALDLS should be recognised as a core, safety-critical component of acute hospital care for people with a learning disability, rather than an optional add-on. In practice, services should be more visibly embedded within routine pathways, with consistent site-based presence, clear referral criteria, early identification through electronic flagging and notification systems, and routine involvement in multidisciplinary planning for complex admissions and procedures. At policy level, ALDLS provision should be recognised within equality and patient safety frameworks as an essential service requiring sustained investment, national minimum configuration standards, adequate staffing, and better-integrated digital systems to support continuity, equitable access, and person-centred care.

12.
arXiv (CS.CL) 2026-06-18

Breaking the Solver Bottleneck: Training Task Generators at the Learnable Frontier

The limiting resource for training agents via reinforcement learning (RL) is increasingly frontier task supply: valid, solvable tasks just difficult enough to train the current model. As reasoning and agentic models improve, fixed task distributions saturate, while naive synthetic generation yields tasks that are trivial, impossible, or ill-posed. Training a task generator with RL to optimize validity and learnability can address this bottleneck, but direct optimization requires repeated solver rollouts per candidate. For software-engineering (SWE) tasks, a single rollout can take tens of minutes; solver-in-the-loop generator training is intractable. We introduce PROPEL, a solver-amortized framework for training task generators at the targeted solve rate. PROPEL trains a lightweight activation probe on a one-time labeled corpus of generated tasks and solver outcomes. The probe predicts target-solver pass rate from a frozen generator reference model and serves as a proxy for solve rate during generator optimization, reducing generator evaluation to a single forward pass. Across math, code, and software-engineering at multiple model scales, PROPEL shifts generation toward the targeted solve rate: for coding, tasks generated at the learnable frontier increase from $10.1\% \rightarrow 20.0\%$ for a Qwen2.5-3B-Instruct solver and from $5.3\% \rightarrow 12.6\%$ for a Qwen2.5-7B-Instruct solver. For SWE, PROPEL increases the share of generations at the targeted solve rate from $9.8\% \rightarrow 19.6\%$ for Qwen3.5-27B on repositories not seen during training of probe and generator.

13.
arXiv (CS.AI) 2026-06-15

The Silent Cost of Artificial Intelligence Assistance: A Theory of Autonomy Surrender, the Recovery Mechanism, and the Restoration of Human Agency

arXiv:2606.13962v1 Announce Type: cross Abstract: The integration of artificial intelligence into human decision-making environments has introduced a previously undertheorized cost: the gradual surrender of human autonomy in exchange for access to information and computational assistance. Building on the Human Identity and Autonomy Gap (HIAG) framework, this paper advances a theoretical model of autonomy surrender as a measurable, cumulative process driven by cognitive bandwidth depletion. The model proposes three interacting mechanisms: the silent cost of AI assistance, in which autonomy is transferred incrementally and without awareness; the surrender threshold, beyond which reclaiming autonomous function becomes cognitively and psychologically difficult; and the recovery mechanism, which establishes the design obligation and the ethical responsibility accompanying deliberate human re-assumption of control. The paper argues that human re-entry into the decision loop is not a passive option but an active cognitive event requiring intentional bandwidth restoration. The design of AI systems must incorporate structured re-entry pathways, here termed recovery mechanisms, that preserve human agency while appropriately distributing responsibility. The model further predicts a terminal state, here termed preference inversion, in which functional dependence on AI assistance is experienced not as a deficit but as a preference, transforming the restoration of autonomy from a design problem into a cultural and political one. Implications are drawn for AI system design, governance frameworks, and human factors research.

14.
arXiv (quant-ph) 2026-06-16

Enhanced Sensitivity near a Quantum Exceptional Point in the Absence of Engineered Dissipation

arXiv:2606.16060v1 Announce Type: new Abstract: Non-Hermitian systems exhibit phenomena absent from Hermitian systems, including exceptional points (EPs), at which two or more eigenvectors coalesce. Conventional implementations rely on gain and loss, which strongly limit quantum coherence. Here, following a proposal by Wang and Clerk (PRA 2019), we realize a closed four-mode quantum system that emulates the dynamics of a PT dimer - two coupled resonators with balanced gain and loss - without engineered dissipation. The four modes are implemented as harmonics of a superconducting coplanar-waveguide resonator, with parametric couplings engineered using a current-pumped SNAIL. We use this device as a sensor for small variations in the PT dimer coupling strength. From signal-to-noise-ratio measurements, we observe enhanced sensitivity near the EP in a non-quantum-limited regime.

15.
arXiv (CS.AI) 2026-06-16

LatentGym: A Testbed For Cross-Task Experiential Learning With Controllable Latent Structure

arXiv:2606.15306v1 Announce Type: cross Abstract: We envision continually learning agentic systems that become more useful over time: as they encounter sequences of related tasks, they should infer the hidden structure shared across those tasks and use it to improve future decisions. This cross-task experiential learning capability is pivotal in domains such as personalization and interactive assistance, but existing training/evaluation frameworks do not provide shared, controllable latent structures and cannot measure whether or why agents improve. We introduce LatentGym: a controllable suite in which each environment is organized around a ground-truth latent variable governing the structure across tasks. Our construction yields metrics that separate exploration (whether the agent's actions gather information about the latent) from exploitation (whether the agent uses what it has gathered). We demonstrate our suite on empirical studies addressing three questions: how and why frontier models fail to adapt across related tasks; whether post-training on related task sequences improves general cross-task adaptation, and where those gains come from; and how design choices such as inter-task feedback shape training dynamics and generalization. Together, these results establish a controlled foundation for studying how LLM agents learn from experience across tasks, and for designing agents that adapt more reliably in sequential, personalized, and interactive settings.

16.
arXiv (CS.AI) 2026-06-16

Automating Low-Risk Code Review at Meta: RADAR, Risk Calibration, and Review Efficiency

arXiv:2605.30208v2 Announce Type: replace-cross Abstract: AI-assisted coding tools have altered software production. At Meta, significant lines of code per human-landed diff grew by 105.9% year over year and per-developer diff volume rose 51%, with agentic AI responsible for over 80% of that growth. Meanwhile, the share of diffs receiving timely review has declined, exposing a widening gap between code supply and reviewer bandwidth. We ask three questions that progress from feasibility through calibration to impact: (1) can risk-stratified automation operate at scale across diverse organizations, (2) how does tuning the risk threshold affect the trade-off between automation yield and safety, and (3) to what extent does automated review reduce end-to-end latency for AI-generated changes? We deployed RADAR (Risk Aware Diff Auto Review), a multi-stage funnel that classifies each diff by authorship and source type, applies eligibility gates, static heuristics, a machine-learned Diff Risk Score, LLM-based Automated Code Review, and deterministic validation before landing qualifying changes. We evaluate RADAR through telemetry covering 535K+ RADAR-reviewed diffs, observational before-after comparisons for policy changes, and difference-in-differences analysis of efficiency outcomes. RADAR has reviewed 535K+ diffs and landed 331K+. Relaxing the Diff Risk Score threshold from the 25th to the 50th percentile increased the approve rate to 60.31%. The revert rate for RADAR-reviewed diffs is 1/3 that of non-RADAR diffs, and the Production Incident rate is 1/50 that of non-RADAR diffs. RADAR reduces median time to close by over 330% and median diff review wall time by 35%. Risk-aware layered automation can materially reduce review bottlenecks created by AI-driven code growth without compromising production safety.

17.
arXiv (CS.AI) 2026-06-15

Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models

arXiv:2606.07157v2 Announce Type: replace Abstract: Many efforts to ensure frontier AI models are safe rely on monitoring their chain-of-thought (CoT) reasoning. If models become able to perform sufficiently complex reasoning internally, without explicit thinking tokens, this would undermine such oversight. We measure how well frontier models reason without CoT across a suite of over 30,000 questions spanning 43 benchmarks in domains including math, coding, puzzles, causality, theory-of-mind, and strategic reasoning. To compare models against humans, we estimate the $50\%$-task-completion time horizon (TH): the human time required for tasks a model completes with $50\%$ success rate. We complement this with a $50\%$ reasoning token horizon: the minimum number of o3-mini reasoning tokens needed for tasks a model solves with $50\%$ success rate. We find that the no-CoT $50\%$ TH of frontier models has been doubling roughly every year over the past six years, with GPT-5.5's TH reaching over 3 minutes and reasoning token horizon exceeding 1,500 tokens. Our median estimates predict that frontier no-CoT THs could exceed 7 minutes by 2028, and 25 minutes by 2030, though these projections carry substantial uncertainty. We recommend frontier developers track this explicitly.

18.
arXiv (CS.CV) 2026-06-16

Dynamic Black-hole Emission Tomography with Physics-informed Neural Fields

With the success of static black-hole imaging, the next frontier is the dynamic and 3D imaging of black holes. Recovering the dynamic 3D gas near a black hole would reveal previously-unseen parts of the universe and inform new physics models. However, only sparse radio measurements from a single viewpoint are possible, making the dynamic 3D reconstruction problem significantly ill-posed. Previously, BH-NeRF addressed the ill-posed problem by assuming Keplerian dynamics of the gas, but this assumption breaks down near the black hole, where the strong gravitational pull of the black hole and increased electromagnetic activity complicate fluid dynamics. To overcome the restrictive assumptions of BH-NeRF, we propose PI-DEF, a physics-informed approach that uses differentiable neural rendering to fit a 4D (time + 3D) emissivity field given EHT measurements. Our approach jointly reconstructs the 3D velocity field with the 4D emissivity field and enforces the velocity as a soft constraint on the dynamics of the emissivity. In experiments on simulated data, we find significantly improved reconstruction accuracy over both BH-NeRF and a physics-agnostic approach. We demonstrate how our method may be used to estimate other physics parameters of the black hole, such as its spin.

19.
arXiv (CS.AI) 2026-06-12

Muse Spark Safety & Preparedness Report

arXiv:2606.12429v1 Announce Type: cross Abstract: Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refusal across a range of benchmarks related to hazardous workflows in chemistry and biology. We therefore release Muse Spark as the underlying model of Meta AI.

20.
arXiv (CS.CL) 2026-06-17

Evaluating Large Language Models Abilities for Addressee, Turn-change, and Next Speaker Prediction in Meetings

We investigate turn-taking in multimodal multi-party conversations using large language models (LLMs). We construct an evaluation framework for three tasks: addressee detection, turn-change prediction, and next speaker prediction. We compare supervised models trained for these tasks, text-based LLMs, multimodal LLMs (MM-LLMs), and human subjects. Experiments on the AMI corpus showed that LLMs outperformed supervised models and humans in next speaker prediction, despite not being trained on the target domain and without access to audio or visual information. An MM-LLM performed better than text-based LLMs on addressee detection and turn-change prediction but remained below human performance, indicating difficulty leveraging raw audio-visual signals. Ablation analyses revealed that conversational context was critical, particularly for next speaker prediction. We observed that human and LLM prediction patterns were similar, and intervals with frequent turn changes were difficult for both.

21.
arXiv (CS.CL) 2026-06-16

AuAu: A Benchmark for Auditing Authoritarian Alignment in Large Language Models

The worldwide surge of authoritarianism, combined with the increasing central role in users' everyday lives, raises the question of to what extent specific models exhibit or promote authoritarian attitudes and characteristics. We introduce AuAu, a comprehensive benchmark that aims to assess the risk of LLMs generating responses with authoritarian tendencies. This benchmark combines three evaluation approaches: (i) psychometric questions from an extensive pool of 15 human validated instruments; (ii) contextual behavior vignettes probing intended actions in concrete situations; and (iii) responses to realistic user prompts. Unlike prior work, AuAu evaluates not only a general closeness towards authoritarianism but also the established sub-concepts Authoritarian Aggression, Authoritarian Submission, and Conventionalism. Evaluating 17 models from China, the EU, Russia, and the USA, we find that all tested models exhibit substantial authoritarian response rates under the psychometric evaluation, though rates drop significantly in increasingly more realistic downstream task. We further find that an authoritarian system prompt easily manipulates 15 out of 17 models to promote increased authoritarianism. Our results underscore the need for continued, systematic auditing of LLM-based AI systems to detect and ultimately mitigate undesired authoritarian tendencies in generated output. Our code and data are available at: https://github.com/andreaseinwiller/AuAu

22.
bioRxiv (Bioinfo) 2026-06-16

RetroMol: Parsing a shared encoding from natural products and their biosynthetic gene clusters

Natural products such as polyketides and nonribosomal peptides (NRPs) are important sources of bioactive compounds, including many antibiotics. Many of them are assembled by modular enzyme complexes and further modified and diversified by tailoring reactions encoded by biosynthetic gene clusters (BGCs). Although natural products and their coding BGCs describe different data modalities of the same biochemical process, a unified language to jointly describe their biochemistry is lacking. Here we introduce a sequence-based representation of the core biosynthesis of modular natural products, which we call primary sequences, that bridges chemical structures and BGCs. We also present RetroMol, an algorithm that parses either natural product structures or their encoding BGCs into their primary sequences of natural product building blocks. RetroMol allows for similarity scoring between natural products and BGCs, enabling the retrieval of compounds, BGCs, and a combination of the two, based on their biosynthetic similarity. This can, for instance, be used to retrieve biosynthetically similar but structurally dissimilar compounds, or link natural products to candidate coding BGCs in large experimental datasets. We demonstrate the latter by rediscovering the nocardichelin B BGC as a proof of principle. We also exemplify the utility of biosynthetic similarity by showing various pairs of biosynthetically similar compounds with low structural similarity. Together, these results establish primary sequences as a shared biosynthetic encoding for natural product comparison and BGC prioritization.

23.
arXiv (CS.LG) 2026-06-17

MiniFool – Physics-Constraint-Aware Minimizer-Based Adversarial Attacks in Deep Neural Networks

arXiv:2511.01352v2 Announce Type: replace Abstract: In this paper, we present a new algorithm, MiniFool, that implements physics-inspired adversarial attacks for testing neural network-based classification tasks in particle and astroparticle physics. While we initially developed the algorithm for the search for astrophysical tau neutrinos with the IceCube Neutrino Observatory, we apply it to further data from other science domains, thus demonstrating its general applicability. Here, we apply the algorithm to the well-known MNIST data set and furthermore, to Open Data data from the CMS experiment at the Large Hadron Collider. The algorithm is based on minimizing a cost function that combines a $\chi^2$ based test-statistic with the deviation from the desired target score. The test statistic quantifies the probability of the perturbations applied to the data based on the experimental uncertainties. For our studied use cases, we find that the likelihood of a flipped classification differs for both the initially correctly and incorrectly classified events. When testing changes of the classifications as a function of an attack parameter that scales the experimental uncertainties, the robustness of the network decision can be quantified. Furthermore, this allows testing the robustness of the classification of unlabeled experimental data.

24.
arXiv (CS.LG) 2026-06-11

Bergson: An Open Source Library for Data Attribution

arXiv:2606.11660v1 Announce Type: new Abstract: Data attribution is a promising field in interpretability that aims to explain model behavior through the influence of its training data, with applications including debugging undesirable model behavior and training dataset curation. However, significant engineering effort is required to perform it at scale, and many cutting edge techniques lack open-source tooling and support. Bergson is an open source library that aims to enable faster progress in the field by providing a host of techniques that scale to very large language models and pre-training datasets. The library natively supports on-disk gradient stores and multi-node distributed training, and provides quality of life tools for researchers. Finally, we introduce the first open-source implementations of three leading data attribution methods: MAGIC, SOURCE, and TrackStar. The library is available at https://github.com/EleutherAI/bergson .

25.
medRxiv (Medicine) 2026-06-16

Optimal Clinical Trials Platform for Progressive Multiple Sclerosis (OCTOPUS): protocol for an international, multi-arm, multi-stage, platform, randomized controlled, double-blind, phase 3 clinical trial.

Introduction Current treatments for multiple sclerosis (MS) do not address the pathological processes of neurodegeneration and chronic demyelination. This, coupled with the significant challenges of translating promising phase 2 results to phase 3 trial success, highlights the need for more efficient trial designs, such as platform multi-arm multi-stage (MAMS) trial approaches. MAMS trials have demonstrated success in areas such as oncology and infectious diseases. They are typified by a statistically robust core trial design that allows the addition of further treatment arms and utilisation of interim outcome analyses at pre-defined timepoints, to determine whether to terminate a treatment arm early or proceed to the final outcome analysis. To address the challenges in progressive multiple sclerosis (PMS) treatment discovery, the Optimal Clinical Trials Platform for PMS (OCTOPUS) trial was developed. It currently utilises MRI whole-brain atrophy as its interim outcome measure and the clinically relevant composite Expanded Disability Status Scale Plus (EDSS-Plus) as its final outcome measure. A rigorous and systematic drug selection process that assessed preclinical in vitro and animal model evidence, along with additional human data, led to the prioritisation of R/S-alpha lipoic acid (R/S-ALA) and metformin for testing against placebo, targeting pathobiological mechanisms relevant to PMS. All participants will be eligible to receive the current standard of care, including disease-modifying treatments (DMTs). Method and analysis OCTOPUS will be a multi-centre, randomised, placebo-controlled, double-blind, phase 3, MAMS trial of participants aged 25 to 70 years (inclusive) with PMS and an EDSS score of 4.0 to 8.0 (inclusive). Steady progression must be the major cause of increasing disability rather than relapse in the preceding 2 years. In the trial s first candidate drug cycle, participants will be allocated to R/S-ALA, metformin, or placebo in a 1:1:1 ratio. Cycle 1 active treatments will start as R/S-ALA 600 mg once daily, increased after 4 weeks to 600 mg twice daily, or metformin 1 g once daily, increased after 4 weeks to 1 g twice daily. The trial will be multinational, with participation from 28 hospitals across the UK and 10 hospitals in Australia. Clinician-reported measures will include: the EDSS-Plus and the individual components: EDSS, Timed 25 Foot Walk (T25FW); 9 Hole Peg Test (9HPT); Symbol Digit Modalities Test (SDMT); Sloan Low Contrast Visual Acuity (SLCVA); and Relapse assessment. Patient-reported outcomes include MS specific walking, fatigue, pain, and impact scales. We will include a health economic analysis. Analysis stage 1 will require randomisation of 125 participants per arm and utilise MRI percentage brain volume change (PBVC) with the Structural Image Evaluation using Normalisation of Atrophy (SIENA) technique from baseline to 78 weeks. A positive outcome in analysis stage 1 will detect a 0.15% per year whole brain atrophy difference with a one-sided alpha of 0.35 and power of 95%, ensuring a low probability of erroneously rejecting a treatment arm at this stage. Any arms that show a positive effect will proceed to final analysis stage 2. Analysis stage 2 will require 600 participants per arm. Participants included in stage 1 will also be included in the stage 2. Analysis stage 2 will evaluate time to 6-month confirmed disability progression in the EDSS-Plus, in order to detect a 25% hazard ratio reduction with 90% power and an alpha of 0.05. Assuming one treatment arm proceeds to analysis stage 2, the trial will recruit approximately 1,200 participants and last about 6 years. This is approximately two-thirds the size and half the duration of separately conducted two-arm phase 2 and 3 trials. Ethics and dissemination The protocol was approved by the London Hampstead REC (22/LO/0622). This manuscript is based on protocol version 8.0, 28th August 2025. The findings of this trial will be disseminated through peer-reviewed publications and conference presentations. There will be a close communication strategy developed with the UK MS Society (MSS) and full patient and public involvement and engagement (PPIE). Trial registration ISRCTN: 14048364 EudraCT number: 2021-003034-37 CTA 20363/0445 IRAS number: 1003943 Secondary identifying numbers: ND001, CPMS 54274 Strengths and limitations - The OCTOPUS trial will be the first platform multi-arm multi-stage phase 3 trial in PMS, offering the potential to significantly expedite clinical trial processes with advantages in cost- and time-efficiency, focusing specifically on the poorly treated pathobiological processes of chronic neurodegeneration and demyelination - It will begin by assessing two promising drug candidates, immediate-release metformin and R/S-ALA, and will expand over the duration of the trial to include more drug arms under the same trial master protocol - The flexible and statistically robust trial design means that several components of the design (such as the early analysis stage 1 interim outcome) can be updated in line with evolving scientific knowledge - It will ultimately be the largest ever investigator-initiated phase 3 trial in PMS - It will include a range of national and international trial sites, including neuroscience centres and district general hospitals - It will have a high inclusion limit for age (up to 70 years) and disability (up to EDSS 8.0) - Several components (the telephone EDSS and virtual patient-reported outcome measures) will be amenable to remote collection increasing inclusivity and thus addressing public and participant suggestions, while minimising the risk of missing data - The main challenges in this trial design are the statistical and methodological complexity involved in design and implementation, and interpretation of interim trial results. Conclusion The trial launched cycle 1 in January 2023. Analysis stage 1 recruitment of 375 participants was achieved in November 2024, enabling planned interim analysis stage 1 to be conducted by late 2026 (Figure 1). On the 1st of June 2026, in the UK, 24 sites are active with a further 4 in set-up as part of stage 2, and in the Australian extension, Platform Adaptive Trial for Remyelination and Neuroprotection in Multiple Sclerosis (PLATYPUS), 1 site is active, with 9 additional sites in set-up.